Healthcare Expenditures Associated With Implementing an External Facilitation Program to Increase the Prescribing of Medications Used for the Treatment of Opioid Use Disorder Among Veterans

Abstract

Background:

To address the opioid use disorder (OUD) public health crisis, the ADvancing Pharmacological Treatments for OUD (ADaPT-OUD) external facilitation randomized trial was conducted in 8 intervention and 27 matched control low-performing Veterans Health Administration (VHA) facilities to increase the prescribing of medications for OUD (MOUD). Facilities were considered low-performers if they were in the bottom quartile of the facility ratio of Veterans with OUD who received MOUD. The objective of this analysis was to evaluate the healthcare expenditures of Veterans with OUD who received care in ADaPT-OUD intervention facilities compared to those receiving care in matched control facilities.

Methods:

Difference-in-differences (DID) design was used to compare the overall, outpatient, and inpatient expenditures (extracted from the VHA data warehouse) of Veterans diagnosed with OUD or receiving MOUD between the 2 groups 12 months before and after the intervention.

Results:

A total of 7348 Veterans with a diagnosis of OUD or prescribed MOUD on at least 1 encounter 12 months after ADaPT-OUD intervention at all sites (92.39% male and 83.26% white) were included for analysis. ADaPT-OUD intervention did not have a substantial impact on overall healthcare costs. However, we reported 4% fewer total encounters in the intervention sites (DID, 95% confidence intervals [CI]: 0.96 [0.92–1.00]) compared to the control sites, driven by a decline in non-VA services. Notably, the outpatient psychiatric-related costs were $391 (95% CI: $49-$733) higher per Veteran within the year after the intervention sites received external facilitation compared to control sites.

Conclusions:

Veterans at intervention sites with an OUD history had higher outpatient psychiatric-related costs, which could be explained by increased access to optimal mental health services at VHA. Improving access to OUD treatment at VA may lead to more coordinated and comprehensive treatment of both OUD and other associated mental health and physical comorbidities.

Introduction

According to the United States (US) National Institute on Drug Abuse, opioid-related overdose deaths increased from 49 860 in 2019 to 81 806 in 2022, a relative increase of 64%.¹ Evidence-based treatments like naloxone distribution for harm reduction and medications for opioid use disorder (MOUD) provide key opportunities to mitigate the rising number of opioid-related deaths.² Medication treatment is the gold standard for the treatment of OUD.^2,3 However, the adoption of MOUD prescribing faces challenges due to system-, clinician-, and patient-level barriers, specifically the need to acquire an X-waiver to prescribe buprenorphine and gaps in knowledge about the effectiveness of MOUD treatment.^4–7 Although the X-waiver requirement was eliminated on December 29, 2023 by President Biden’s signing of the Consolidated Appropriations Act of 2023,⁸ developing optimal strategies to improve access to MOUD remains an important priority for healthcare systems.

Implementing evidence-based treatment practices is fraught with challenges due to resource limitations, organizational culture, and support and prioritization of the implementation process.⁹ Efforts to bridge these gaps include external facilitation strategies, which aim to encourage healthcare systems to adopt evidence-based treatment programs and provide a unique strategy to address these implementation issues.^10–12 The ADvancing Pharmacological Treatments for OUD (ADaPT-OUD), an external facilitation study, commenced in 2017 to bolster MOUD prescribing for patients with OUD at low-performing US Veterans Health Administration (VHA) medical centers.¹³ The ADaPT-OUD study focused primarily on employing external facilitation to increase uptake of buprenorphine prescribing (due to its lower regulatory burden compared to methadone) and injectable naltrexone for cases where buprenorphine is not an acceptable alternative for a particular patient.¹³ The ADaPT-OUD external facilitation intervention resulted in an improvement in MOUD provision, particularly among facilities with low baseline MOUD levels. Using a per-protocol analysis (which excluded 1 facility that dropped out immediately before the intervention phase), the intervention led to a significant overall improvement in MOUD provision (4.7%, 95% confidence intervals [CI]: 1.1%−7.9%) and the number of patients prescribed buprenorphine (23.0 patients, 95% CI: 6.1–39.5).^7,14 Due to these findings, VHA launched the Stepped Care for Opioid Use Disorder Train the Trainer Initiative, replicating ADaPT-OUD study’s facilitation implementation strategies, to improve access to MOUD in primary care, mental health, and pain clinics throughout VHA.^15,16 Despite the efficacy of these external facilitation approaches, decision-makers capable of adopting and implementing this strategy will need to determine how to optimize the benefits of the intervention given their current budget constraints.

According to the Quadruple Aim Framework, healthcare systems must simultaneously pursue 4 dimensions of health system performance: improve patient experience of care, improve the health of populations, reduce the per capita costs of health care, and improve healthcare provider well-being and productivity.¹⁷ As an evidence-based implementation strategy, facilitation for MOUD could potentially improve the overall health of the OUD population by increasing access to MOUD, enhancing patients’ quality of life by reducing the risk of opioid overdose and death, and diminishing long-term costs associated with opioid-related treatments and consequences.

Decision-makers will need to consider the economic consequences of evidence-based treatment programs and balance the value generated with the opportunity costs of forgoing the next best alternative.^17,18 Understanding implementation costs is crucial for the adoption of evidence-based treatment, encompassing accounting costs for capital equipment and labor required for the supply, delivery, and provision of the program.¹⁹ More importantly, information in the form of downstream costs is necessary for decision-makers to realize the long-term cost offsets associated with implementing an evidence-based treatment program and reconciling their budget.^18,20

Understanding the patterns of expenditures before and after an implementation intervention informs the budget impact of implementing an evidence-based treatment program for OUD. Therefore, we sought to evaluate the downstream costs of implementing an external facilitation program (ADaPT-OUD intervention) to increase the adoption of MOUD at low-performing VHA facilities. The primary objective of this study was to compare total healthcare expenditures among Veterans with a diagnosis of OUD in low-performing VHA facilities that received and did not receive external facilitation. The secondary objective was to evaluate the impact of external facilitation on Veterans’ healthcare expenditures among Veterans with OUD in non-VHA facilities paid for by VHA.

Methods

Study Design

A budget impact analysis using a fixed effects difference-in-differences (DID) model was performed to compare the patient-level downstream healthcare expenditures on Veterans diagnosed with OUD 12 months before and after implementation of the external facilitation program (ADaPT-OUD intervention). The implementation or index date was defined as the first ADaPT-OUD-related site visit at the 8 intervention sites. The site visit date ranged between March 12, 2018 and June 25, 2019. Control sites were paired with an intervention site and adopted the same index date. This study served as a follow-up study to a cluster, randomized controlled trial comparing the efficacy of the ADaPT-OUD intervention on VHA facilities’ MOUD prescribing rates versus VHA facilities with no facilitation.^7,13,14 External facilitation encompassed local barrier/facilitator assessment, establishment of a local implementation team, site visits for action planning and training/education, monthly coaching calls, cross-facility quarterly calls, and consultations.¹⁴ We followed the guidance from Wagner et al on best practices for budget impact analysis in implementation research.²⁰ The research protocol was reviewed and approved by the ********** and ********* Institutional Review Boards and was conducted in accordance with all applicable federal regulations.

Participated Cohort

VHA facilities with the lowest quartile of MOUD provision for patients diagnosed with OUD (as of October 2017) were randomized to intervention (8 sites) and control (27 sites) groups.¹³ These 35 facilities were stratified by prescribing rate (low, <15% and high, 15%−21%) and the number of patients diagnosed with OUD but without a MOUD prescription (“actionable patients”: low, <472 and high, ≥472 patients).¹³ The purpose of the ADaPT-OUD intervention was to increase the provision of MOUD as measured by the percentage of patients diagnosed with OUD and prescribed an MOUD. Therefore, all Veterans who had a history of OUD diagnosis or MOUD prescribed within 12 months after the index date at one of the study sites were included for analysis. We used this criterion to restrict our sample to Veterans who had used VHA services during the study period. Veterans with a history of OUD were identified using the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM). Veterans who had at least 1 documented MOUD prescription (buprenorphine, naltrexone [injection], or methadone [measured as an orderable item for an outpatient visit to VA clinical stop code for Opioid Treatment Program]) 12 months after the index date in the assigned sites were identified by pharmacy data files from the VHA Corporate Data Warehouse (CDW). To conduct the DID design, Veterans were included in the study if they had at least 1 healthcare encounter 12 months before the index date. Details of the diagnostic codes and pharmacy files are available in Table S1.

Outcomes and Data Sources

Our primary outcomes included total inpatient and outpatient expenditures (costs and resource utilization) provided or paid for by VHA; this included both VHA- and non-VHA-related care (also known as community care). The annual expenditures were estimated for each individual Veteran 12 months before and after the index date using administrative data from the VHA CDW. These data included the VHA outpatient National Patient Care Database, inpatient transfer files for utilization, and the VHA Managerial Cost Accounting files for cost data.^21,22 We classified total costs and resource utilization (encounters for services in settings such as ambulatory, emergency, diagnostic, lab, home health, inpatient, and virtual encounters) into several subcategories. Outpatient care expenditures were classified into 5 categories (Medical/Surgical, Psychiatry/Mental Health, Substance Use Disorder [SUD], pharmacy, and other [eg, dental, dialysis, diagnostic, adult daycare, ancillary services, home care, etc]). Inpatient care expenditures were classified into 4 categories (eg, Medical/Surgical, Psychiatry/Mental Health, SUD-related, and other [eg, domiciliary, nursing home, and rehabilitation]). Other healthcare expenditures administered by non-VHA facilities (secondary objective) and paid by the VHA were obtained from payments reported in VHA Fee Basis files.

Independent Variables

We included patient-level factors, such as sex, age (<65 and ≥65 years old), race (White, Black, other, and unknown), ethnicity (Hispanic/Latino, not Hispanic/Latino, and unknown), marital status (married, single, divorced/separated, widow, and unknown), secondary insurance (public, private, and other/unknown), number of the Elixhauser conditions,^23,24 and military service-connected disability status. Clinical conditions (alcohol use disorder [AUD], nicotine use disorder, post-traumatic stress disorder, suicide ideation or attempts, depression, OUD, dementia, anxiety, and history of any hematology/oncology visits) were based on diagnoses recorded in the year prior to the index date. In addition, the cohort was evaluated for the treatment of mental health and psychiatric conditions that were based on pharmaceutical prescription history, which included antidepressants, antipsychotics, SUD other than opioids, benzodiazepines, and sedatives/hypnotics.

Statistical Analyses

Characteristics of the Veteran cohort at the index date were compared using a chi-square test. Healthcare expenditure comparisons between the groups were presented as the mean difference and standard error of the mean (SEM). A fixed effect estimator in the DID approach was used to estimate patient-level changes in healthcare expenditures 12 months before and after the index date relative to the control sites.²⁵ For cost outcomes, we constructed linear regression models, with a clustered sandwich estimator for the variance, which allows for intragroup correlation with each VA medical station, to estimate the average differences with corresponding 95% CI between patients in the intervention and control sites. For resource utilization outcomes, we constructed negative binomial regression models to estimate the incidence rate ratio (IRR) of healthcare utilization with corresponding 95% CI between patients in the intervention and control sites. Statistical significance was defined as a 2-tailed alpha level <5%. All analyses were performed using Stata SE version 18 (Stata Corp, LLC, College Station, TX, USA).

Results

Demographics and Clinical Characteristics

A total of 7348 Veterans were included for analysis with 1351 (18.39%) Veterans located at intervention sites and 5997 (81.61%) Veterans located at control sites (Figure 1). The Veteran population receiving care at an intervention site had a higher proportion who were over 65 years old (25.04% vs 19.61%), of Hispanic or Latino descent (9.7% vs 3.02%), and had higher rates of other SUD (81.91% vs 77.08%) and AUD (38.7% vs 31.64%) diagnoses compared to the Veteran population at control sites (Table 1). In addition, a higher proportion of the Veterans receiving care at intervention sites had ≥3 Elixhauser conditions at baseline, compared to those who received care at control sites (55.44% vs 49.89%, respectively). In contrast, a higher proportion of Veterans receiving care at control sites were in rural or island communities compared to those at intervention sites (39.96% vs 26.80%).

Figure 1.

The flow chart of the ADaPT healthcare expenditures study. To evaluate the ADaPT program, we included Veterans who had at least 1 OUD/MOUD visit within 12 months after the index date in assigned sites.

Abbreviations: ADaPT, ADvancing Pharmacological Treatments for Opioid Use Disorder; MOUD, medications for opioid use disorder; OUD, opioid use disorder.

Table 1.

Baseline Characteristics of Patients Assigned to the ADaPT-OUD or Control Group.

	Total (N = 7348)	Control (N = 5997)	ADaPT (N = 1351)
	N (%)	%	%	P value
Male	6789 (92.39%)	92.55%	91.71%	.295
Age (≥65)	1514 (20.61%)	19.61%	25.04%	<.001
Race				<.001
White	6118 (83.26%)	84.28%	78.76%
Black	841 (11.45%)	10.86%	14.06%
Other	122 (1.66%)	1.55%	2.15%
Unknown	267 (3.63%)	3.32%	5.03%
Ethnicity				<.001
Hispanic/Latino	312 (4.25%)	3.02%	9.70%
Not Hispanic or Latino	6966 (94.80%)	96.10%	89.05%
Unknown	70 (0.95%)	0.88%	1.26%
Marital status				<.001
Married	2330 (31.73%)	32.35%	28.96%
Single	1588 (21.62%)	21.69%	21.33%
Divorced/separated	3176 (43.25%)	42.94%	44.59%
Widow	207 (2.82%)	2.64%	3.63%
Unknown	43 (0.59%)	0.38%	1.48%
Secondary insurancea				.039
Public	2582 (35.14%)	34.57%	37.68%
Private	700 (9.53%)	9.82%	8.22%
Other/unknown	4066 (55.33%)	55.61%	54.11%
Geographic region (rural/insular)	2752 (37.54%)	39.96%	26.8%	<.001
VA service-connected disability (%)				.237
<50	1322 (17.99%)	17.76%	19.02%
≥50	3614 (49.18%)	49.64%	47.15%
Unknown	2412 (32.83%)	32.60%	33.83%
Elixhauser comorbidities number				<.001
0	373 (5.08%)	5.45%	3.40%
1	1604 (21.83%)	22.29%	19.76%
2	1630 (22.18%)	22.36%	21.39%
≥3	3741 (50.91%)	49.89%	55.44%
Alcohol use disorder	2415 (32.94%)	31.64%	38.70%	<.001
Nicotine use disorder	469 (6.40%)	5.97%	8.30%	.002
Drug use disorder	5716 (77.97%)	77.08%	81.91%	<.001
PTSD	2980 (40.65%)	40.54%	41.14%	.684
Suicide ideation or attempt	815 (11.12%)	10.97%	11.79%	.387
Depression	3380 (46.11%)	46.07%	46.26%	.902
OUD	5324 (72.62%)	71.95%	75.61%	.006
Dementia	90 (1.23%)	1.15%	1.56%	.224
Anxiety	2134 (29.11%)	28.79%	30.54%	.200
Medications
Antidepressants	5419 (75.36%)	75.29%	75.66%	.778
Opioids	4326 (60.16%)	59.82%	61.64%	.221
Antipsychotics	2259 (31.41%)	31.29%	31.95%	.640
Stimulants	254 (3.53%)	3.63%	3.09%	.334
Hypnotics (no benzodiazepine)	1290 (17.94%)	17.91%	18.09%	.877
Benzodiazepine	1809 (25.16%)	24.73%	27.05%	.078
Hematology/oncology visits	42 (0.57%)	0.62%	0.37%	.277

Only 72.62% (5324 out of 7348) of subjects had a history of OUD before the index date (baseline), indicating ~30% of new OUD diagnosed after the index date or received medication for OUD without an ICD-10 OUD diagnosis.

Abbreviations: PTSD, post-traumatic stress disorder; ADaPT-OUD, ADvancing Pharmacological Treatments for Opioid Use Disorder; ICD-10, International Classification of Diseases, Tenth Revision; VHA, Veterans Health Administration.

^aPublic insurance includes Medicare, Medicaid insurance; private insurance includes major medical/HMO/PPO/champion/indemnity. Other insurance includes only dental and/or only prescriptions. While all VHA patients are eligible for care through the VHA, many also carry secondary insurance, such as Medicare, Medicaid, private insurance, or other forms of coverage.

Primary Objective

Total Healthcare Costs and Utilization (VHA and Non-VHA)

Table 2 summarizes the average healthcare costs and utilization among the Veterans with OUD who received care at intervention (ADaPT-OUD group) and control sites 12 months before and after the index date. The total average healthcare costs (which included VHA and non-VHA costs) were reduced from $47 578 (SEM 1702) to $44 248 (SEM 1636) in the 12 months before and after the implementation date in ADaPT-OUD intervention sites. In the control sites, the total average healthcare costs were reduced from $39 397 (SEM 686) to $36 090 (SEM 621) 12 months before and after the index date.

Table 2.

Health Care Costs for Patients in the ADaPT and Control Groups, 12 Months Before and After ADaPT Initiation.

	12 Months before	12 Months after	12 Months before	12 Months after
	Control (N = 5997)	Control (N = 5997)	ADaPT (N = 1351)	ADaPT (N = 1351)
	Mean (SEM)	Mean (SEM)	Mean (SEM)	Mean (SEM)
Health care costs
VHA inpatient costs
Medical/surgical	$3877 (306)	$3157 (177)	$4754 (544)	$5148 (609)
Psychiatry/mental health	$6763 (297)	$4818 (249)	$8482 (726)	$6550 (597)
SUD	$1117 (72)	$210 (30)	$1315 (235)	$1 (1)
Other	$6438 (297)	$6660 (302)	$7620 (749)	$8149 (811)
Total VHA inpatient	$18 195 (572)	$14 845 (492)	$22 170 (1378)	$19 848 (1332)
VHA outpatient costs
Medical/surgical	$4092 (75)	$4093 (77)	$4955 (199)	$5396 (231)
Psychiatry/mental	$3950 (81)	$3972 (82)	$4604 (176)	$5016 (212)
SUD	$2235 (56)	$2294 (58)	$2733 (137)	$2674 (147)
Pharmacy	$3699 (100)	$3807 (96)	$3944 (219)	$3836 (182)
Other	$4427 (83)	$4688 (90)	$5579 (251)	$5510 (207)
Total VHA outpatient	$18 403 (230)	$18 854 (237)	$21 815 (584)	$22 431 (581)
Non-VHA costs
Inpatient	$1714 (96)	$1358 (95)	$2083 (238)	$1049 (188)
Outpatient	$1086 (46)	$1033 (52)	$1510 (136)	$919 (94)
Total
VHA and non-VHA outpatient	$19 489 (241)	$19 888 (248)	$23 325 (617)	$23 350 (600)
VHA and non-VHA inpatient	$19 909 (594)	$16 202 (511)	$24 253 (1443)	$20 898 (1369)
Total costs	$39 397 (686)	$36 090 (621)	$47 578 (1702)	$44 248 (1636)
Health care utilization
VHA inpatient encounters
Medical/surgical	0.23 (0.01)	0.20 (0.01)	0.28 (0.02)	0.24 (0.02)
Psychiatry/mental health	0.38 (0.02)	0.29 (0.01)	0.38 (0.02)	0.32 (0.02)
SUD	0.06 (0.00)	0.01 (0.00)	0.04 (0.01)	0.00 (0.00)
Other	0.17 (0.01)	0.19 (0.01)	0.16 (0.01)	0.19 (0.01)
Total inpatient	0.84 (0.02)	0.69 (0.02)	0.86 (0.04)	0.74 (0.04)
VHA outpatient encounters
Medical/surgical	10.76 (0.16)	9.7 (0.14)	11.32 (0.34)	10.73 (0.34)
Psychiatry/mental health	13.5 (0.26)	11.76 (0.25)	15.84 (0.62)	14.2 (0.60)
SUD	11.62 (0.26)	8.92 (0.21)	13.77 (0.66)	10.78 (0.54)
Other, VHA	25.88 (0.39)	19.17 (0.27)	26.72 (0.84)	19.57 (0.56)
Total, outpatient	90.38 (0.88)	78.27 (0.75)	96.98 (2.01)	84.67 (1.74)
Non-VHA encounters
Inpatient	0.21 (0.01)	0.31 (0.04)	0.23 (0.03)	0.3 (0.08)
Outpatient	8.88 (0.52)	8.87 (0.49)	13.99 (1.52)	6.14 (0.74)
Total encounters
VHA and non-VHA outpatient	99.26 (1.03)	87.14 (0.91)	110.97 (2.55)	90.81 (1.98)
VHA and non-VHA inpatient	1.05 (0.03)	1 (0.04)	1.09 (0.06)	1.05 (0.09)
Total	100.31 (1.04)	88.14 (0.91)	112.06 (2.56)	91.85 (2.00)
Inpatient LOS (days)
Medical/surgical	0.94 (0.06)	0.74 (0.04)	1.15 (0.13)	1.03 (0.12)
Psychiatry/mental	4.29 (0.23)	2.41 (0.14)	5 (0.51)	2.85 (0.28)
SUD	1.4 (0.09)	0.26 (0.03)	1.42 (0.24)	0 (0.00)
Other, VHA	9.81 (0.43)	8.4 (0.36)	8.79 (0.81)	8.64 (0.77)
Total, VHA	1.5 (0.13)	2.01 (0.30)	3.16 (1.19)	1.87 (0.45)
Total, non-VHA	16.44 (0.54)	11.8 (0.43)	16.37 (1.12)	12.52 (0.90)
LOS, VHA, and non-VHA	17.95 (0.57)	13.81 (0.53)	19.53 (1.64)	14.39 (1.04)

Abbreviations: ADaPT, ADvancing Pharmacological Treatments for Opioid Use Disorder; SUD, substance use disorder; SEM, standard error of the mean; VHA, Veteran Health Administration; LOS, length of stay.

In the DID analysis, there were no significant differences in the total healthcare costs between the ADaPT-OUD and control groups 12 months before and after the index date; however, significant differences in total resource utilization were reported (Figure 2A and B and Table S2A–C). The total average healthcare utilization was 112.06 (SEM 2.56) and 91.85 (SEM 2.00) encounters 12 months before and after the index date, respectively, among Veterans who received care at intervention sites. The total average healthcare encounters were 100.31 (SEM 1.04) and 88.14 (SEM 0.91) encounters 12 months before and after the index date, respectively, among Veterans at control sites. Veterans at intervention sites were associated with a 4% reduction in total healthcare encounters compared to Veterans at control sites 12 months before and after the index date (DID, 95% CI: 0.96 [0.92–1.00]; Figure 2B).

Figure 2.

The DID of total healthcare costs (A) and a total utilization IRR (B) between ADaPT and control groups, 12 months before and after intervention.

Abbreviations: ADaPT, ADvancing Pharmacological Treatments for Opioid Use Disorder; IRR, incidence rate ratio; NVA, non Veterans affairs; DID, difference-in-differences.

Medical/Surgical Costs and Utilization in the VHA Facility

In the inpatient setting, we did not observe differences in the Medical/Surgical costs or utilization before and after the index date between the groups (Figure 3A and B). However, the difference in outpatient Medical/Surgical costs before and after the index date was $440 higher for Veterans in the ADaPT-OUD group compared to the control group (DID, 95% CI: $440 [$153-$727]; Figure 3C). We did not observe any significant DID in outpatient Medical/Surgical utilization (Figure 3D).

Figure 3.

The DID healthcare costs (A, C) and utilization IRR (B, D) between ADaPT and control groups, 12 months before and after intervention, by care subcategories in the inpatient and outpatient setting.

Abbreviations: ADaPT, ADvancing Pharmacological Treatments for Opioid Use Disorder; IRR, incidence rate ratio; NVA, non Veterans affairs; DID, difference-in-differences.

The Psychiatric Services-Related Costs and Utilization in the VHA Facility

In the inpatient setting, we did not observe any psychiatric service costs or utilization differences before and after the index date between the groups (Figure 3A and B). However, the difference in outpatient psychiatric service costs before and after the index date was $391 higher for Veterans in the ADaPT-OUD group compared to the control group (DID, 95% CI: $391 [$49-$733]; Figure 3C). We did not observe any significant DID in outpatient psychiatric utilization (Figure 3D).

The Substance Use-Related Costs and Utilization in the VHA Facility

While we did not observe any significant DID in the inpatient substance use treatment costs, the inpatient substance use utilization was 93% lower in the ADaPT-OUD group compared to the control group 12 months before and after the index date (DID, IRR, 95% CI: 0.07 [0.01–0.56]; Figure 3A and B). Accordingly, the length of stay was 91% less in the ADaPT-OUD group compared to the control group 12 months before and after the intervention (DID, IRR, 95% CI: 0.09 [0.01–0.64]; Figure 2C). In the outpatient setting, we did not observe any significant costs or utilization differences before and after the index date between the groups (Figure 3C and D).

Pharmacy Healthcare Costs in the VHA Facility

Veterans in the ADaPT-OUD group had an increase in pharmacy-related costs before and after the index date when compared to Veterans in the control group (DID, 95% CI: $86 [−$585 to $757]), but this difference was not statistically significant (Table S2A).

Other Healthcare Costs and Utilization in the VA Facility

In the inpatient setting, the change in costs associated with other-related healthcare before and after the index date was not significantly different between the groups (DID, 95% CI: $364 (−$1436 to $2163]; Figure 3A). Similarly, in the outpatient setting, the DID in costs associated with other-related healthcare was not significantly different (DID, 95% CI: −$329 [−$687 to $30]; Figure 3C). Further, there were no significant DID in the incident rate in terms of consuming other-related healthcare resources for the outpatient (DID, IRR, 95% CI: 0.98 [0.93–1.03]) and inpatient settings (DID, IRR, 95% CI: 1.08 [0.88–1.32]; Figure 3B and D).

Secondary Objective: Healthcare Costs and Utilization in Non-VHA Facilities

The healthcare costs decreased in both non-VHA inpatient (DID, 95% CI: −$ 678 [−$1236 to −$119]) and outpatient (DID, 95% CI: −$539 [−$1067 to −$10]) settings in the ADaPT-OUD group compared to the control group 12 months after the index date (Figure 3A and C). In addition, the non-VHA outpatient healthcare utilization was 23% less in the ADaPT-OUD group compared to the control group 12 months after the implementation date (DID, IRR, 95% CI: 0.77 [0.68–0.88]; Figure 3D).

Discussion

We sought to compare the patterns in healthcare expenditures between Veterans with OUD who were at sites that did and did not implement external facilitation (ADaPT-OUD intervention) to improve MOUD prescribing. Healthcare expenditures 12 months before and after the study index date varied across the groups. In our study, Veterans receiving care at ADaPT-OUD facilitation intervention sites had significantly greater reductions in total healthcare encounters compared to control sites before and after the implementation index date, but there were no differences in total costs. Conversely, Veterans at the intervention sites had a significantly greater increase in VHA-related outpatient costs (eg, acute medicine and psychiatric services) before and after the intervention compared to Veterans at the control sites, which was offset by a significant reduction in non-VHA inpatient and outpatient costs.

Although the DID findings indicated that Veterans receiving care at ADaPT-OUD facilitation intervention sites experienced significantly greater reductions in total healthcare encounters compared to control sites (with no differences in total costs), we observed a reduction in healthcare expenditures among Veterans with OUD at both intervention and control sites after the index date. The ADaPT-OUD intervention was implemented about 1 year after the MISSION Act of 2018, which expanded Veterans’ access to non-VHA community care.²⁶ Prior studies have associated non-VHA care with higher costs and lower value.^27,28 It is possible that Veterans initially rushed to utilize non-VHA care before the index date, leading to higher pre-intervention healthcare expenditures. Over time, the reliance on non-VHA care may have diminished, reducing overall healthcare expenditures during the follow-up year as Veterans received care within the VA system.

Implementing ADaPT-OUD facilitation intervention at low-performing VHA facilities may have led to more coordinated and comprehensive treatment of both OUD and other associated psychiatric and physical comorbidities. We observed increased Medical/Surgical and Psychiatric-related costs in the ADaPT-OUD group compared to the control group before and after the index period; however, we did not observe an increase in encounters. We speculate that the ADaPT-OUD intervention may have led to a change in practice where less-costly suboptimal care was substituted with costly optimal care. Given that these Veterans were located at low-performing sites when it comes to MOUD dispensing, we speculate that these sites also had a high incidence of low-value services prior to the index date,^27,28 which are responsible for increased acute Medical/Surgical healthcare costs after the index date without a change in utilization. SUD-related care is substantial and may explain the increased costs without an increase in encounters.²⁹ This pattern has been previously reported for patients diagnosed with OUD who incurred greater costs after an opioid-related overdose without a significant change in their outpatient encounters (eg, longer [45 or 60 minutes], and group psychotherapy).³⁰ Further work will be needed to sufficiently understand and evaluate the mechanisms behind these healthcare expenditure patterns among Veterans with OUD.

We also reported that the ADaPT-OUD group was associated with a 93% reduction in inpatient substance use utilization; however, the absolute reduction may not be clinically meaningful.³¹ These differences represented small absolute changes in visits for the ADaPT-OUD (0.04 and 0.00 visits/Veteran) and control (0.06 and 0.01 visits/Veteran) groups before and after the index date, respectively. Inpatient detox/rehabilitation treatment centers are relatively more costly than other OUD treatment settings, including behavioral health and outpatient services.³² The lower inpatient substance use utilization after the ADaPT-OUD intervention was consistent with previous findings indicating a reduced risk of all-cause, opioid-related, and alcohol-related mortality, as well as a lower risk of injury, poisoning, and suicide following use with MOUD.^33,34 Still, the lack of significant change in inpatient and outpatient costs secondary to substance use in the ADaPT-OUD intervention could be attributed to the low utilization of inpatient services at low-performing VHA facilities (both intervention and control groups) and requires further evaluation. The ADaPT-OUD intervention was conducted in low-performing VHA sites with few Veterans with OUD at the study sites receiving MOUD during the intervention period. This relatively low uptake likely contributed to the absence of a significant impact on overall pharmacy costs and utilization. In addition, it is possible that the small proportion of Veterans receiving MOUD may have been offset by reductions in the use of other medications (eg, for managing withdrawal symptoms or non-MOUD treatments for OUD), leading to a net neutral effect on pharmacy costs.

We observed a greater reduction in non-VHA costs and utilization with the ADaPT-OUD intervention group compared to the control group before and after the index period, which may have been prompted by a need to receive optimal SUD-related care at VHA. Veterans with OUD can receive care through VHA-purchased care at non-VHA facilities, a provision significantly expanded by the Veterans Choice Act of 2014 and the MISSION Act of 2018.^26,35 We speculate that making these services more readily available at VHA sites may have encouraged Veterans to return to VHA for MOUD and other associated mental and physical comorbidities, which is evident with the decrease in non-VHA outpatient and inpatient encounters. Moreover, several reports suggest VHA-related care was associated with higher quality care (ie, cognitive behavioral therapy for other mental health conditions) and lower costs.^36,37 Chan et al²⁷ reported that patients who were ≥65 years old and received care at VHA had a 46% reduction in mortality and a 21% reduction in spending compared to Veterans who received care in the community (non-VHA).

Although our study did not evaluate the implementation cost of ADaPT-OUD, which may be an important consideration for stakeholders, a previous study by Garcia et al reported that the cost of implementing the ADaPT-OUD facilitation intervention may be affordable depending on the budget of the healthcare system.³⁸ According to the authors, the composition of the stakeholders (eg, leaders, providers, staff) was a major factor in the overall cost of implementation due to their varying salaries. Decision-makers planning to implement ADaPT-OUD facilitation will need to take this into account when developing their budget.

Limitations

Due to the nature of our study design, several limitations deserve discussion. First, there were significant covariate imbalances between the groups. The ADaPT-OUD study used a clustered-randomized design where randomization occurred at the VHA facility level; however, our analysis was at the patient level, which resulted in an unbalanced risk profile between participants in the ADaPT-OUD and control groups.³⁹ To mitigate this problem, we used a fixed effects model, which takes into consideration the within-subject effect assuming that the observed and unobserved baseline covariates are time-invariant.²⁵ Second, our study may be subject to selection bias, as Veterans who received MOUD could differ significantly from those who did not in terms of baseline health status or healthcare engagement. The ADaPT-OUD intervention was implemented at VHA medical facilities identified as underperforming in MOUD management (MOUD prescription rates of <21%).¹³ Restricting the cohort to only those Veterans with an OUD diagnosis who initiated MOUD after the index date would exclude a substantial portion of the population with OUD. As such, this approach may not fully capture the broader impact of the intervention on Veterans with an OUD diagnosis. In a previous study, VHA clinicians who received the required X-waiver prescribed buprenorphine below their capacity indicating that other barriers to access beyond credentialing may be responsible.⁴⁰ Thus, any expected increase in MOUD prescribing due to the intervention would be attenuated due to clinicians prescribing below their capacity. Next, given that community care is generally less effective and more costly than VHA-related care,²⁷ it is important for us to understand the interaction between the effects of the ADaPT-OUD intervention from those of community care on healthcare expenditures. We attempted to explore this by reporting on the changes in non-VHA care separately from VHA-related care; non-VHA-related costs and resource utilization decreased in the intervention group compared to the control group with a corresponding increase in VHA-related costs and resource utilization suggesting a potential substitution effect. Future investigations will be needed to verify these observations. Lastly, Veterans have twice the mortality risk of drug overdose thereby having a different baseline risk profile compared to the general US population, which may limit generalizability to other large, integrated healthcare systems similar to VHA (eg, Kaiser).³⁴ Conclusions drawn from our analysis should take these differences into consideration.

Conclusions

The decline in healthcare expenditures during the follow-up year was observed in both the intervention and control groups, though the DID analysis indicated that Veterans who were at VHA sites with the ADaPT-OUD facilitation intervention had a 4-percentage point reduction in total healthcare utilization compared to Veterans in the control sites from 12 months before to 12 months after the index date. Significant reduction in non-VHA-related expenditures (costs and resource utilization) may have contributed to the lower overall healthcare utilization (VHA and non-VHA) among Veterans who were at VHA sites with the ADaPT-OUD facilitation intervention, but this was balanced by significant increases in VHA-provided outpatient Medical/Surgical and Psychiatric costs. Decision-makers may find the results of this study informative when making decisions about implementing an external facilitation program to promote MOUD prescribing, but they may need to anticipate a trade-off between paying for care within and outside the network.